Species-Area Relationship
Intuitively we would all expect to find more species in, say, California than in my back yard in Pasadena. What's unexpected is that this implied relationship between area and number of species apparently follows a universal law. In particular, scientists have found empirically and repeatedly no matter what group of organisms studied, no matter what part of the earth, no matter what overall scale the number of species roughly doubles each time you increase the area by a factor of ten.
More precisely, species grow as a power of area: if you plot species versus area on log-log paper you invariably get a straight line that is, if you plot the logarithm of species versus the logarithm of area, a common and highly useful technique used in diverse fields throughout science, from seismology to electrical transmission lines. Strictly speaking, however, one could argue that this is not a universal law that would require the line being identical for all species and areas. In fact, the slope of the line differs from place to place and from one group of organisms to another. But the form of the relationship is universal; that is, the line is always straight.
I will be the first to confess that it is somewhat difficult to create a plausible dataset that doesn't fit a straight line on log-log paper! For example, any polynomial relationship will present a pretty darned straight line in log space above a certain threshold especially given the wild uncertainties that seemingly plague all aspects of ecology. But to put this in perspective, take a minute to imagine what other laws might have been chosen. One might reasonably expect, for instance, the number of species to be a combination of competing factors, each factor being dominant over different scales. Small islands, for example, are limited in habitat diversity, while large islands have a full complement and therefore are limited by an entirely different mechanism. Thus one might expect species to drop off steeply below a certain size. But, in fact, the same law seems to apply simultaneously to islands a mile across and to islands hundreds of miles across.
Historically, this law was best studied in island groups, where boundaries are clear and populations are sometimes quite stable, but it turns out it applies equally well to any so-called habitat islands (such as lakes or isolated mountains), and even to arbitrary areas of the mainland. For example, scientists have studied forests fragmented by logging, lakes left behind by retreating ice-age glaciers, and even urban parks!
Typically the power falls between about 0.25 and 0.33, that is, species grow as approximately the cube-root or fourth-root of the area. (Isn't that the power you'd have chosen?) Here's a random sample of data I pulled off the web showing how closely diverse groups of organisms match this law:[1]
Some datasets are astonishingly faithful to this law over wide ranges of areas -- take, for instance, the species of plants on the Galapagos (see inset plot). This plot represents over 4 orders of magnitude in area! A similar plot of the flora of California spans 5 orders of magnitude. [Where??]
Exceptions that Prove the Rule
Like all so-called laws in complex sciences, there are exceptions. The power can actually vary from as flat as 0.1 to as steep as 0.72 according to one site I found on the web:[3] [Tom and I are convinced this is bogus data.]
Other scientists prefer an exponential law instead of a power law. See, for example, the impressive analysis Jared Diamond and Ernst Mayr did on the Solomon Islands, spanning fully six orders of magnitude in area the largest island was a million times larger than the smallest! [4] Such an exponential law, as it turns out, matches our power law fairly well, except at the extreme lower end (islands smaller than 1 square mile).
What's It Good For?
One of the primary arguments against the species-area law is that it is not predictive. One must take a fair amount of data and estimate two values -- both the power and a constant multiplier. You cannot, the argument goes, a priori predict how many species one should find in a given region without first measuring how many species are present in dozens of very similar regions.
True enough, but there are other perfectly valid predictions one can make with it. For example, one can measure the power law of species present in existing forests of a given region, and from that estimate how many species will be lost if forests are reduced due to logging, fires, disease, development, etc. In fact, SAR they would have an acronym for it, wouldn't they? is a standard technique used in ecological management throughout the world.
For example, knowing nothing at all, we can make a back-of-the-envelope estimate about the effects of rainforest destruction in the Amazon. If 90% of the forest is destroyed, then we can immediately say with some confidence that half of the species will be lost. [NOTE]
Why Does It Work?
Why? Ah, that's the question, isn't it? Numerous factors have been identified and studied, but not surprisingly no one factor seems to be able to account for all the observations.
By far the most celebrated theory to date is the Equilibrium Model of Island Biogeography. Roughly speaking it balances the rate of colonization of incoming species against the rate of extinction due to population pressure. If one can estimate these two rates, based on such diverse factors as competition, resource scarcity, distance from the mainland and other islands, and so on, then in theory one can predict the form the species-area curve will take in a given instance. In practice, it seems this is bit overly-optimistic. [7]
Other studies show that habitat diversity by itself offers a better explanation, the theory presumably being that the larger the number and diversity of ecological niches, the larger the number of species that can cohabit the region. [8]
Edge effects can also make a big difference. Forest next to clear-cuts, for example, receives too much sun, greater wind damage, and responds by changing the structure of the understory and overstory dramatically. While this creates new habitat increasing habitat diversity at the same time it reduces the effective size of the preexisting forest habitat. It has been claimed that the area of the Amazon affected by deforestation actually more than doubles if you take edge effects into account! [9]
Over time climate change and even evolution itself will change the species. For example, as continental glaciers receded at the end of the Pleistocene, some regions, such as the Great Smoky Mountains in the southern Appalachians, provided better refuge for displaced species than surrounding areas, and thus ended up with abnormally high diversity. While, on the other hand, other regions might turn inhospitable, and we must wait for evolution to expand the few surviving species into the newly available, novel niches. Urban zones offer a poignant example of the latter effect: evolution will eventually populate our cities with a diverse fauna and flora, no doubt wonderfully adapted for the harsh conditions, but unfortunately for us it will most likely take many lifetimes for nature to catch up.